Utilizing machine learning algorithms for cardiovascular disease prediction: "Detailed analysis based on medical parameters"

Abstract

Among the most prevalent and dangerous ailments impacting human health are cardiovascular diseases (CVDs). Early diagnosis may help avoid or lessen CVDs, thereby lowering death rates. Several clinical methods have already been deployed for diagnosing and treating CVD. However, one interesting approach is to use Machine Learning (ML) approaches to identify risk characteristics. The suggested model uses a variety of ML approaches to accurately forecast cardiac disease. Initially, the CVD dataset is collected and trained in the Python tool. The null and duplicate records are removed in the data preprocessing stage. Moreover, extracts relevant information from the dataset using feature extraction. Inter Quartile Range (IQR) is used in AdaBoost and Gradient Boosting to identify continuously distributed outliers in data. Moreover, 16 ML classifiers are employed to accurately forecast the CVD disease. Compared with other approaches, the AdaBoost and Gradient Boosting approach gained better results of 96 %. The developed model dataset is trained and tested with k-fold testing. GridSearchCV and the results are visualized using the SHAP tool. The designed technique enhances the CVD prediction system using several MLs.

Keywords: Cardiovascular disease; Feature extraction; Interquartile range; Machine learning; Prediction; SHAP; k-fold Testing.